The AI alignment properties of agents which would be interesting to a range of principals trying to solve AI alignment. For example: